Approaches for Achieving Broadband Achromatic Phase Shifts for Visible Nulling Coronagraphy Matthew R. Bolcar' and Richard G. Lyon NASA Goddard Space Flight Center, 8800 Greenbelt Rd., Greenbelt, MD 20771 ABSTRACT Visible nulling coronagraphy is one of the few approaches to the direct detection and characterization of Jovian and Terrestrial exoplanets that works with segmented aperture telescopes. Jovian and Terrestrial planets require at least 10-9 and 10-10 image plane contrasts, respectively, within the spectral bandpass and thus require a nearly achromatic "'-phase difference between the arms of the interferometer. An achromatic "'-phase shift can be achieved by several techniques, including sequential angled thick glass plates of varying dispersive materials, distributed thin-film multilayer coatings, and techniques that leverage the polarization-dependent phase shift of total-internal reflections. Herein we describe two such techniques: sequential thick glass plates and Fresnel rhomb prisms. A viable technique must achieve the achromatic phase shift while simultaneously minimizing the intensity difference, chromatic beam spread and polarization variation between each arm. In this paper we describe the above techniques and report on efforts to design, model, fabricate, align the trades associated with each technique that will lead to an implementations of the most promising one in Goddard's Visible Nulling Coronagraph (VNC). Keywords: coronagraphy, interferometry, achromatic phase shift 1. INTRODUCTION The direct observation of Terrestrial planets in the visible bandwidth would allow for spectroscopic analysis to determine planetary composition, as well as the presence of water and the possibility to support life. To achieve direct detection, an instrument must be capable of high-<:ontrast imaging, or differentiating the 10 orders-of-magnitude difference between the host star's diffracted light and that reflected by an orbiting Terrestrial planet. Furthermore, as more and more telescopes are being designed with segmented or sparse apertures, the direct-detection technique should be compatible with these architectures I". The visible nulling coronagraph (VNC) is a direct detection technique that achieves all of these requirements. As shown in Figure 1, the VNC uses a symmetric Mach-Zehnder interferometer to suppress the starlight at an inner working angle (IWA) of -2.VD. A MEMS segmented deformable mirror (OM) in one arm of the interferomeler provides wavefront control. When coupled with a fiber-bundle array, simultaneous amplitude and phase control is achieved with a single OM. Furthermore, the fiber-bundle array spatially filters the wavefront, relaxing high spatial­ frequency wavefront requirements on the system. The segmented nature of the OM also makes the VNC compatible with segmented and sparse aperture systems. As the name implies, the VNC operates in the visible bandwidth regime. In order to enable spectroscopy of the detected planets, the VNC must also accommodate broad bandwidths, ideally spanning the band from 400 nm to 700 nm. This implies that an achromatic phase shift of 1t radians must be introduced between the arms of the interferomeler. Many techniques for achieving an achromatic phase shift have been developed for white-light interferometry. For example, a series of dispersive plates can be designed to balance the optical path length (OPL) at several wavelengths (similar to the design of achromatic doublet lenses)"'). Passing a beam through focus"', or sequential reflections from mirrored surfaces can also introduce a phase shift', at the cost of remapping the pupil of the optical system. Thin-film coatings can be used to achieve broadband performance of optics'. Recently, a series of Fresnel rhomb prisms, augmented by sub-wavelength gratings, have been used to achieve an achromatic phase shift in the infrared' . •matthew .bolcar@nasagov M1 :" 51,ea/ "' :' Piston Lyot MechaL'i;sm :. :.... .. ...........r- .... Meehanis.' $top-2 =;'" . ~ _ '" M6 ~ - ~ - 'j -----_. ~."......-- ;'. M4 Ann-2 ~ Input Lighl 8S-1 c'O">6~ '~"'l/~, ~9 ..... v ~">If <T Figure I - Optical layout of the visible nulling coronagraph, Light enters from "'to telescope at the lower left and is split and recombined by two matched beamsplitters. Light reflecting off the lit beamsplitter traverses two flats (MI and M2) and reflects otT a MEMS hex-packed segmented deformabLe mirror. This Light both reflects and transmits through the 2"" beamsplitter and is combined with the light reflecting otTlL.ts M4, M5 and M6. There are two output channels labeled as the bright object sensor (BOS) used for fine pointing and wavefront control, and the science (SCI) channe~ where an in· focus image of star system without the starlight is collected. While all of these techniques have been developed previously, none have to date achieved the extreme performance requirements for high-contrast imaging al broadband, visible wavelengths. We report here recent successes in two separate designs 10 achieve a broadband visible achromatic ,,-phase shift for the VNC. 2. REQUIREMENTS OF ACHROMATIC PHASE SHIFTER 2.1 Contrast Dependence on Wavefront Error The achievable contrast of the VNC depends on a number of instrumental properties and ultimately determines several performance requirements on the achromatic phase shifter (APS). Before discussing those requirements, however, it is illustrative to discuss the dependence of contrast on something more straightforward: the difference in the wavefront error between the arms of the interferometer (for a more detailed discussion of the operation of the VNC, the reader is directed to references I and 9). The noise-free image irradiances in the bright and dark focal planes of the VNC are given by 1B(e)=~I1PSF,(e)+~PSF; (e)+~1II2 Re{ ASF,(e) ASF,' (e)} (I) ID(e)=~I1PSF,(e)+~PSF,(e)-~M2 Re{ ASF,(e) ASF;' (e)} • where I.(fJ) and Io(fJ) represent the bright and dark channel output images, respectively, e is the angular variable representing the focal plane projected on the sky, II and I, are the integrated intensities in each arm of the interferometer such that II + I, = 1'ID" ASFt!.. fJ) is the unaberrated, diffraction-limited complex amplitude spread function, ASF. is the aberrated (phase & amplitude) complex amplitude spread function and PSF, and PSF. are the point spread functions given by PSF, (e) = ASF, (e) ASF; (e) (2) pSF,(e) = ASF, (e) ASF; (e) , All of the phase and amplitude aberrations can be ascribed to one ann or the other of the interferometer without loss of generality since it is only the difference between the anns that matters. Using the small angle approximation for the phase error of e'< ~ 1+ it/> , we can expand the dark channel equation for I D( 0, which leads to the following expression for image-plane contrast: c=.!.L~(1I:W,), PSF,(6-6,), (3) I"", A. where e, is the location of the planet and Wo is the amplitude of the wavefront error at the spatial frequency that corresponds to that location in the image plane. If the mean wavefront error is zero (equivalent to the piston difference between the interferometer arms being zero) then the average contrast is (4) The brightness of speckles is exponentially distributed such that its mean is equal to the standard deviation of its intensity. Since we desire to set the requirements for the VNC based on high-confidence statistics we require (5) where CFT'gh' is the flight requirement contrast limit. This insures that the flight contrast limit will be met better than 99"10 of the time. Solving Eq. (4) for the wavefront error and computing the RMS value gives (6) per spatial frequency. 8 For the lab VNC operating at a contrast of C = 10. aLl = 633 om requires I1w'" 0.014 om per spatial frequency. The overall RMS wavefront error (WFE) is obtained by integrating I1w over all spatial frequencies of interest. Spatial frequencies of interest are limited for the VNC to what is controllable by the segmented deformable mirror. The deformable mirror has 163 active segments, each with 3 control degrees of freedom (OOF), for a total of 489 control OOF in all. To achieve contrast of 10" with 489 OOF at). = 633 om requires I1w:S 0.247 om. This is the requirement on the RMS difference of the wavefront error between the two anns of the interferometer, if all other error sources are considered negligible. 2.2 Contrast Depeodence on the APS Of course, there are other error sources that are non-negligible, including intensity variations due to coating imperfections, polarization variations and errors due to a rmite spectral bandpass. For each of these terms, an analysis similar to the one performed in Section 2.1 will yield requirements on how well the APS must control these error sources. We can re-express the bright- and dark-channel intensities from Eq. (1) in terms of the variance of eacb of the error terms: 1 u ' u' U' 11:' (<lA )'l} IB(6) = \+~41~ 3 {\- [11:' ( ; )+ l~+--:-+ 48;::- J (7) 1 2 u u' U' 11:' LU In(9)- 4 11: ~ ' +_1 +~+-(- )'] 1+ ~ lei' [ ( A. ) 16 4 48 A., where Uw is the RMS wavefront error as computed in Section 2.1 , u/ is the variance of the fractional intensity difference between the arms of the interferometer, u/ is the variance of the difference in polarization-vector rotation between the arms in units ofradians, A< is the spectral bandwidth of the system, and we've made use of the asymptotic form of the PSF: PSF(9) =[1+ ~41~3 r (8) The last term in each expression in Eq. (7) is due to spectral leakage from the path length difference between the two arms of the interferometer. Ideally, the path length difference is half of the central wavelength, -<0, to yield a phase difference of (21</ Ao)OPL = 1< where OPL is the optical path length difference. At wavelengths different than Ao, the phase difference does not result in perfect destructive interference and light is leaked through, lowering the contrast.